Markov type and superreflexivity
From OpenAI Aftermath
Markov type and superreflexivity: Ball's Markov type is a metric inequality that bounds how far a stationary reversible Markov chain, mapped into a normed space, can travel in t steps. A Banach space is superreflexive if it admits an equivalent uniformly convex norm. Naor asked whether a Banach space can have nontrivial Markov type without an equivalent uniformly smooth norm. A manuscript in OpenAI's public mathematics collection (github.com/openai/math), dated 23 September 2026, claims that every real Banach space with Markov type p > 1 is superreflexive, which answers the question negatively and, with the known converse, characterizes superreflexivity by nontrivial Markov type. This page summarizes the published claim and lists publicly available follow-up work. Statements are the respective authors' claims; the publication of a paper or a formal proof is not presented here as independent verification.
Background
- As recounted in the introduction of the manuscript:
- Ball introduced Markov type to study Lipschitz extension. Superreflexivity has metric characterizations in the Ribe programme: Bourgain's by the distortion of finite complete binary trees, another by finite diamond graphs, and Mendel and Naor's by Markov convexity, introduced by Lee, Naor and Peres. Rademacher type alone does not imply it: James constructed nonreflexive spaces of Rademacher type 2.
- Naor, Peres, Schramm and Sheffield showed that uniform smoothness of power type p, 1 < p ≤ 2, implies Markov type p; with Pisier's renorming theorem, every superreflexive space has nontrivial Markov type. Naor asked whether the converse holds, as Question 4 of a survey.
The original result
- Claim. Theorem 1.1 of Nontrivial Markov Type Forces Superreflexivity states that every real Banach space with Markov type p for at least one p > 1 is superreflexive. Together with the known converse, a real Banach space has nontrivial Markov type exactly when it admits an equivalent uniformly convex norm, equivalently an equivalent uniformly smooth norm.
- Method, as described in the proof strategy: arguing by contradiction, the James–Enflo characterization gives a nonreflexive space finitely representable in the given space, from which a spreading norm on finite sequences with contractive merging of consecutive entries is built, a concrete version of the Brunel–Sucheston construction. For each time n a finite reversible walk with unit edge labels is constructed whose sum has norm at least n/24 with probability at least 3/20, using reduced words acting on the line, finite Ramsey averaging and a lattice lift that keeps the chain reversible; this contradicts Markov type p > 1.
- Source (openai/math repository, commit
7ef5c0c): Nontrivial Markov Type Forces Superreflexivity, dated 23 September 2026, 14 pages. - Formalization, as described by the repository. The scope note lean/docs/327.md states that the formalization proves that a real Banach space has nontrivial Markov type exactly when it admits an equivalent uniformly convex norm, hence exactly when it is superreflexive, with consequences for uniformly smooth renormings and the reflexivity of finitely representable spaces, and links the comparator statement MarkovType.lean.
Public follow-up work
| Date | Work | Relation | Public claim | Public status (checked 29 October 2026) |
|---|---|---|---|---|
| 2026-10-06 | Markov type and superreflexivity, an article of the Latent Seminar blog at lowrank.github.io | Reading (exposition) | Explains reversible walks and equivalent norms, finite representability and an ordered norm, the passage from edge labels to actual increments, the contradiction in one estimate, and how Markov type differs from other probabilistic invariants, with outlines of the key steps. | Blog article; it states that its review covers statements, introductions and selected proof mechanisms and is not an independent verification of the complete proofs. |
| 2026-10-14 | Markov type and superreflexivity, a course on Open Mathematics Courses (kokunoyumeto/open-math-courses-public) | Reading (exposition) | Five lessons on the manuscript's proof: Markov type and the reduction to a finite walk, a spreading norm with contractive merging, charts and reversible transport, two levels from reduced words, and the finite walk. The course states that the theorem of James and Enflo is its one named hypothesis and that everything else is proved. | The lessons state that they were written by Claude Opus 5.5 (Anthropic) in October 2026 and self-checked by the writing AI; the course lists its status as author self-checked. |
| 2026-10-19 | AI Proved What?, episode 327 Walks reveal the shape?, remade version of 23 October 2026 | Reading (exposition) | Video explainer of the claim that every real Banach space with Markov type p for some p above 1 is superreflexive, so that, with the known converse, nontrivial Markov type is exactly superreflexivity. The episode states that no relation between the exponents, no explicit constant beyond the proof's own and nothing about general metric spaces is claimed, and that the walks built in the proof are finite but astronomically large. | Web page and video; the site states that its scripts and animations are produced with AI assistance and that the narration is an AI-generated voice. Status record last checked 27 October 2026: posted as a preprint: OpenAI, 14 pages, dated September 23, 2026; under expert review: not yet; peer reviewed: not yet; checked in Lean: the full equivalence. The page also states that the Lean formalization of the full equivalence, including the classical inputs, has 207 files and about 27,000 lines, not run by the site, that three outside projects re-ran it and it was accepted, and that it found no public comment from specialists and no reported error. |
Notes
- The background summarizes the manuscript's own account of earlier work.
- The list covers follow-ups found in public sources up to the date above; it is not a priority or completeness claim.
